[Works fine with hassakuXLIllustrious_v30, konanmixNoobaiXL_v10 and ntrMIXIllustriousXL_xiii is also good] <s>My first LoRa model, done on CivitAI trainer. </s>V4 re-train and lil' update, now closer to original. It tries to recreate Hayabusart's futa OC "Lucca" <a target="_blank" rel="ugc" href="https://x.com/hayabusart/status/1608069838056099848">https://x.com/hayabusart/status/1608069838056099848</a> Current version training set - 13 images. Trigger word: hayLucca Additional words (more in example pics): <span style="background-color:rgb(37, 38, 43);color:rgb(140, 143, 163);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:14px">white skin, green eyes, no pupils, white sclera, mole under eye, mole, orange hair, multicolored hair, short hair, spiked hair, shaved temples, dark temples, ear cuff, ear piercing, piercing, ears, earlobe, thick brows, eyebrows, short brows, eyelashes, eyelids, eyeliner, gigantic breasts, black nails, nail polish, bracelet, black choker, half-closed eyes, white tank top Adetailer is recommended. Upscale by 1.5, Hires steps 0, Denoising strength 0.6, Hires CFG Scale 5 Adetailer Settings: ADetailer steps 32, ADetailer CFG scale 5, Inpaint width &amp; height 1024, Inpaint denoising strength 0.3, ADetailer CLIP skip 2 Training dataset:<a target="_blank" rel="ugc" href="https://files.catbox.moe/0fan6t.zip"> Available in model download. (Use it for your own LoRA if you want)</a>

Lucca [Hayabusart] - Illustrious

V1 - Fail. V2 - Somewhat works, bit aggressive in pushing unwanted details, poses and stuff. &lt;- Final one. V3 - V2 but worse.  V4 and V4.5 - Fail.

Files with SHA256: 83e588a7e809abb071e3c0d8a7fd6e7025e06898abf2f3917242957f3af40390